Presentation (CTI)

OBJECTIVES: Adjustment for treatment effect modifiers (TEM) and prognostic factors (PF) is essential for robust and unbiased population-adjusted indirect treatment comparisons (PAIC) such as matching-adjusted indirect comparisons and multi-level network meta-regression. However, guidance on selecting these variables remains limited. This study aimed to describe current practices for TEM and PF selection in National Institute for Health and Care Excellence (NICE) technology appraisals (TA) and summarise evidence assessment group (EAG) feedback.
METHODS: A targeted review of publicly available NICE TAs published since January 2024 was conducted to identify appraisals that utilised PAIC methodology. For each TA, data were extracted from the company submission, EAG report, and final appraisal documents. Extracted information included chosen TEMs and PFs, the rationale and process for selection, PAIC methodology, and EAG and committee critiques relating to variable adjustment and selection. Findings were summarised descriptively.
RESULTS: Across included TAs, TEMs and PFs were consistently included in PAICs. Variable selection was typically based on literature reviews, subgroup analyses, and clinical expert input, with a lack of systematic or structured approach across TAs to identify TEMs and PFs. EAG feedback frequently highlighted concerns relating to incomplete or inconsistent adjustment, such as insufficient matching for key variables, and criticism of justification for inclusion and exclusion of variables, leading to potential violations of underlying PAIC assumptions. These concerns contributed to uncertainty in treatment effect estimates and, in some cases, reduced confidence in the comparative evidence results.
CONCLUSIONS: Selection of TEMs and PFs in NICE submissions remains heterogeneous and EAGs frequently raise concerns about incomplete or inconsistent adjustments, impacting confidence in treatment effect estimates. These findings highlight the need for clearer guidance and standardisation processes to enhance the credibility of PAICs and support more robust decision making in health technology assessment.